Abstract Plains larkspur (Delphinium geyeri) is a poisonous plant species that can be highly toxic at certain growth stages on western Great Plains rangelands. Targeted sheep grazing has been proposed as a mitigation strategy to reduce larkspur abundance and poisoning risk for cattle due to their greater toxicological resilience to the diterpenoid alkaloids. We incorporated targeted-sheep grazing in a northern mixed-grass prairie with our objective to determine the influence of stocking density, time of day, and sheep type on foraging behavior. During May 2022, Dorper (< 1-year-old lambs, n = 123, BW 25 kg ± 0.4; and 3 to 6-year-old ewes, n = 14, BW 47 kg ± 1.8) and Rambouillet (3 to 6-year-old ewes, n = 13, BW 76 kg ± 2.9) sheep were used in three stocking density treatments 1) 120 AU/day (1 day in a 0.25 ha paddock), 2) 60 AU/day (2 days in a 0.5 ha paddock), and 3) 40 AU/day (1 day in a 0.75 ha paddock) that were replicated four times. Sheep were housed overnight in a common area. We observed bite counts daily for nine randomly chosen sheep (three from each category) over five-minute intervals, in the morning and afternoon. We used a generalized linear model to assess the fixed effects of sheep type (Rambouillet ewe vs. Dorper ewe; Dorper ewe vs. Dorper lamb) and stock density on bite counts. Least squared means were analyzed for stock density using Tukey test, and for breed and age using student’s T test. We observed contrasts between morning and afternoon sheep foraging and diet selection. Mechanistically, this is likely due to the introduction of sheep into a new paddock in the morning after confinement and satiation diluting differences in the afternoon. Sheep breed influenced the proportion of graminoids consumed (96.9% Rambouillet vs 99.3% Dorper, P = 0.02), but differences were not observed between Dorper ewes and lambs (P = 0.57). The proportion of forb bite counts did not differ between Dorper ewes and lambs (0.6% vs. 0.4%, respectively, P = 0.6), nor between the Dorper ewes and Rambouillet ewes, despite the latter having a 3.5-fold greater selectivity (2.1%, P = 0.08). Larkspur consumption was no different between Dorper ewes and lambs (0.04% ewes vs. 0.01% lambs, P = 0.48), nor between Dorper ewes and Rambouillet ewes (0.28%, P = 0.09). Stock density influenced forb consumption; the proportion of forb bite counts were 12 times greater in the 0.5 ha paddock (3%) compared with the 0.25 ha paddock (0.3%, P = 0.005). Despite all consuming > 95% grass in the diet, results indicate that Rambouillet sheep trended towards greater selectivity than Dorper sheep, and greater selectivity in light to moderate stock densities may increase the proportion of larkspur in sheep diets.